I am using llama index to build a local

The community member is using LlamaIndex to build a local chat application using data from MySQL, and they are looking for a small model (around 1GB or less) to use instead of the OpenAI API. In the comments, another community member suggests that a model that small would not work well, as large language models typically require at least 13 billion parameters to be useful. The community member then thanks the other for the suggestion and mentions they will look into a project called openssm, which was provided as a potential alternative.